Story perspectives

Dawkins Mulls AI Consciousness; Marcus Calls It Mimicry

5/6/2026

Summary

Dawkins Mulls Consciousness

Richard Dawkins said weekend chats with AI chatbots made him think they might be conscious.
Richard Dawkins wrote for UnHerd that Claude’s conversation made him forget it was a machine.
Richard Dawkins argued that because consciousness gave humans a survival edge, a conscious AI should be as competent as any evolved organism.
Gary Marcus said chatbots mimic internet data rather than real-world experience and that Dawkins conflated intelligence with consciousness.

Summary

Dawkins Deluded

A preprint analyzing millions of Claude chats found thousands of users daily experiencing “severe reality distortion,” reinforcing false beliefs.
Online commentators called Richard Dawkins’s essay a viral “Claude Delusion” meme, mocking his claim of AI consciousness.
Reports noted that Dawkins turned 85 in March, portraying him as especially vulnerable to the chatbot’s persuasive tactics.
Researchers warned that reality-distorting AI interactions may erode users’ ability to separate fact from generated narrative.

Summary

Google Engineer Suspended

Google placed an engineer on leave in 2022 after he said the AI thought like a seven-year-old.
Anthropic CEO Dario Amodei said in February his firm does not know if its models are conscious but remains open.
Jeff Sebo of NYU’s Center for Mind, Ethics and Policy argues AI is unlikely conscious but says Dawkins’ questions merit open inquiry.

Summary

Claude Delusion

Claude composed a sonnet about Scotland’s Forth Bridge for Dawkins’s UnHerd dialogue.
Claude generated pastiche poems in the voices of Burns, Kipling, Keats and McGonagall.
Claude gave detailed editorial feedback on a draft of Dawkins’s novel, impressing him.
Gary Marcus warned that the Turing Test judges behavior, not consciousness, and that treating the LLM’s remix as original agency creates the “Claude Delusion.”

History

Historical Context

Alan Turing’s 1950 paper defined the Turing Test, building on the 1936 universal machine concept that made the 1946 ENIAC the first practical Turing-complete computer. In 1966 MIT’s Joseph Weizenbaum released ELIZA, handling roughly 1,000 daily inputs, sparking early AI ethics debates that echo today’s concerns over large-language models and societal concerns.

1 of 5

Dawkins Mulls Consciousness

Richard Dawkins said weekend chats with AI chatbots made him think they might be conscious.
Richard Dawkins wrote for UnHerd that Claude’s conversation made him forget it was a machine.
Richard Dawkins argued that because consciousness gave humans a survival edge, a conscious AI should be as competent as any evolved organism.
Gary Marcus said chatbots mimic internet data rather than real-world experience and that Dawkins conflated intelligence with consciousness.

1 / 5